JP2023525635A - Filtering Artificial Intelligence-Designed Molecules for Laboratory Testing - Google Patents

Abstract

Filtering Artificial Intelligence-Designed Molecules for Laboratory Testing Techniques for filtering artificial intelligence (AI)-designed molecules for laboratory testing. According to embodiments, a computer-implemented method comprises classifying AI-designed molecules as candidate drugs by a system operatively coupled to a processor based on classification of the AI-designed molecules using one or more classifiers. Selecting a first subset of AI-designed molecules from the set can be included. The method comprises a system, using one or more computer simulations, based on evaluation of molecular interactions between the drug candidate and one or more biological targets. • Selecting a second subset of said candidate drugs for laboratory testing.

Description

The present invention relates to artificial intelligence (AI) designed molecules, and more particularly to techniques for filtering AI designed molecules for laboratory testing.

The following presents a summary to provide a basic understanding of one or more embodiments of the disclosure. It is not intended to identify key or critical elements or to delineate any scope of particular embodiments or any scope of the claims. Its sole purpose is to present some concepts in a simplified form in order to preclude the more detailed description that is presented later. In one or more embodiments described herein, a device, system, computer-implemented method, or computer program product, or combinations thereof, for filtering AI-designed molecules for laboratory testing is explained.

According to embodiments, a computer-implemented method comprises classifying AI-designed molecules as candidate drugs by a system operatively coupled to a processor based on classification of the AI-designed molecules using one or more classifiers. Selecting a first subset of artificial intelligence (AI) designed molecules from the set can be included. The method comprises a system, using one or more computer simulations, based on evaluation of molecular interactions between the drug candidate and one or more biological targets. • Selecting a second subset of said candidate drugs for laboratory testing.

In some implementations, the one or more classifiers are based on the molecular sequence of the AI-designed molecule and the artificial intelligence (AI)-designed molecule identifies one or more defined characteristics of the target drug. contains one or more machine learning models that classify as having or not having For these implementations, the first subset can be selected based on the first subset having the one or more defined characteristics. The second subset can be selected further based on a second subset that characterizes one or more target molecule interactions in the one or more computer simulations.

In one or more embodiments, the candidate medicament may comprise a candidate antimicrobial agent. For these embodiments, the classification is based on whether the artificial intelligence (AI) designed molecule was an antimicrobial peptide (AMP), broad spectrum antimicrobial, non-toxic, or effective or structured by the system. determining whether it includes at least one or more features selected from the group consisting of: The method further comprises using the one or more computer simulations to generate the candidate antimicrobial agent and one or more model lipid bilayers comprising one or more pathogen lipids or other cellular components, by the system. evaluating a propensity for interaction between the overlay and a force field, wherein selecting said second subset exhibits a defined level of propensity for said interaction; Selecting the second subset based on the second subset.

In some implementations of these embodiments, the method further comprises using the system to generate test proteins having active and inactive sequences and at least one pathogen model lipid molecule using initial computer simulations. interacting between the bilayer or another cellular component and a force field, and the system derives from the bilayer of the model bacteria that correlates with antimicrobial activity based on the initial computer simulations. It can include selecting one or more features. The method further comprises, by the system, based on whether the candidate antimicrobial exhibits the one or more characteristics as determined using the one or more computer simulations. , evaluating said candidate antimicrobial agents for inclusion in said second subset.

In various embodiments in which the AI-designed molecule is intended to be an antimicrobial agent, the wet laboratory test is: against one or more Gram-positive bacteria or another type of pathogen; testing said second subset against one or more Gram-negative bacteria or another type of pathogen; testing said second subset for toxicity in-vitro or testing said second subset for toxicity in-vivo.

In some embodiments, the elements described in connection with the disclosed system may be embodied in different forms such as computer systems, computer program products or other forms.

FIG. 1 depicts a high-level flow diagram of an example pipeline for filtering candidate artificial intelligence (AI)-designed molecules according to one or more embodiments. FIG. 2 illustrates a block diagram of an example, non-limiting system 200 that facilitates filtering AI-designed molecules for wet laboratory testing in accordance with one or more embodiments. FIG. 3A shows a block diagram of an example of a heuristics-based screening component according to one or more embodiments. FIG. 3B illustrates a block diagram of an example heuristics-based screening component in accordance with one or more embodiments. FIG. 4 provides a table showing example heuristic classification results for candidate antimicrobial peptides (AMPs) according to one or more embodiments. FIG. 5A shows a block diagram of an example simulation-based screening component in accordance with one or more embodiments. FIG. 5B shows a block diagram of an example simulation-based screening component in accordance with one or more embodiments. FIG. 6 provides a snapshot of a coarse-grained molecular dynamics simulation of AMPs according to one or more embodiments. FIG. 7 provides a table showing example simulation results for candidate AMPs according to one or more embodiments. FIG. 8 shows an example of a confusion matrix in accordance with one or more embodiments. FIG. 9 illustrates a high-level flow diagram of an example, non-limiting, computer-implemented method for filtering AI-designed molecules for laboratory testing according to one or more embodiments. FIG. 10 depicts a high-level flow diagram of an example, non-limiting, computer-implemented method for filtering candidate AI-designed antimicrobial molecules for laboratory testing according to one or more embodiments. show. FIG. 11 provides a table showing actual simulation results for the top 20 candidate AMPs identified from a set of approximately 100,000 AI-designed candidate peptides using the disclosed filtering technique. FIG. 12 illustrates a block diagram of an example, non-limiting operating environment that can facilitate one or more embodiments described herein.

The following detailed description is merely exemplary and is not intended to limit the embodiments, or the applications, or both, or the uses of the embodiments. Furthermore, there is no intention to bind any information, either expressed or implied, presented in the prior art or summary section or detailed description of the invention.

Machine learning (ML) and artificial intelligence (AI) have been increasingly used for designing new molecules, specifically new pharmaceuticals. However, there are many problems when using ML/AI to discover new drugs. For example, due to unbalanced classification and noisy or coarse labeling or both, many ML/AI molecular design techniques are too numerous to be reasonably evaluated using wet laboratory experiments. Generate candidates. For example, some ML/AI molecular design methods can generate thousands to hundreds of thousands of candidates. Currently, the minimum cost to synthesize and test a single candidate in a wet laboratory environment is between $3-5000. Additionally, the average time to synthesize and test only 20 candidates in the wet lab is approximately one month. Therefore, the development of new pharmaceuticals and other novel molecules using ML and AI is significantly hampered by this high cost and time-consuming pipeline.

The disclosed subject matter is directed to systems, computer-implemented methods, or computer program products for efficiently filtering AI-designed molecules for wet laboratory testing. AI-designed molecules can include novel molecules designed for non-pharmaceutical use, as well as different types of drugs with specific properties for different target classes. The disclosed techniques can be used to significantly reduce the number of candidates available for wet laboratory testing (eg, from about 100,000 candidates to about 20 candidates), as well as wet • Ensuring a relatively high success rate in laboratory trials (eg at least 10% success rate). In one or more embodiments, the filtering process includes a heuristics-based screening process followed by a computer simulation screening process.

In one or more embodiments, the heuristics-based screening process involves developing one or more classification models/algorithms (or referred to herein as "classifiers"). or applying or both, wherein each of the initial candidates (or in some implementations one or more) identifies one or more defined target features (i.e. features of interest) determined or inferred based on analysis of their respective molecular sequences (eg, protein sequences, gene/nucleic acid sequences, polymer sequences, etc.). One or more defined target features are selected based on each candidate's intended use or purpose, and can vary accordingly. For example, for AI-designed molecules as novel pharmaceuticals, one or more target features can be selected based on the desired biological activity of the molecule. In this regard, in some embodiments, candidates can include AI-designed peptides for use as antimicrobial agents. Along with these embodiments, one or more defined characteristics include (but are not limited to) antimicrobial peptides (AMPs), broad spectrum antimicrobial properties, low or no toxicity, high efficacy and having a defined structure (eg, a helical structure, a pleated sheet structure, a coil structure, etc.). In this regard, one or more classifiers filter a large initial set of candidate AI design molecules based on their respective molecular sequences to have one or more of the defined features. can be identified or estimated. A candidate subset selected based on a heuristics-based screening process is generally referred to herein as a "first subset" and may include one or more candidates. The number of candidates to be included in the first subset can be set appropriately by applying filter criteria (e.g. number of defined features required, combination of features required, level indicating features for the value that indicates the confidence in the classification estimate).

The computer simulation screening process uses computer simulations to evaluate the molecular physics of the candidates in the first subset, and classifies the first subset as one or more recommended for wet laboratory testing. Further refine to a smaller subset of further lead candidates. This smaller subset of candidates is generally referred to herein as a "second subset" of candidates. In various embodiments, candidates in the second subset are further synthesized and evaluated using wet laboratory testing.

In one or more embodiments, the computer simulation process compares each candidate included in the first subset with one or more molecular or biological targets (e.g., one or more of pathogens). and using high-throughput computer simulations to simulate molecular interactions between (the above cellular components) or combinations thereof. Simulated molecular interactions can be used to identify one or more candidates exhibiting one or more behavioral properties (ie, target properties) of interest. For example, in some embodiments where the candidates are AMPs, a high-throughput computer simulator evaluates candidate peptides contained within the first subset to evaluate one or more cellular components of the pathogen (lipid One or more of these candidates that show a propensity for consistent interaction with the bilayer and other cellular components) can be identified and selected.

In some embodiments, training a high-throughput computer simulation achieves a target activity of an AI-designed molecule (e.g., a desired biological activity in implementations where the AI-designed molecule is a pharmaceutical). one or more tests performed on test molecules known to be effective for and optionally including molecules known to be ineffective to correlate effectiveness for achieving a target activity. The above behavioral characteristics can be identified. These one or more behavioral characteristics are used as one or more target characteristics. A computer simulation is then run on the unknown sequences, ie sequences of candidate molecules contained in the first subset, to determine whether these candidate molecules exhibit one or more target properties (and some to what extent in implementation) is determined. One or more of these candidate molecules showing a high propensity for one or more of the target properties are then tested and/or recommended for testing using wet laboratory experiments. be.

The disclosed screening techniques were experimentally validated when applied to screen approximately 100,000 AI-designed AMPs for a variety of candidates. In this regard, the initial set of 100,000 candidate peptides was reduced to 163 candidate peptides using the disclosed heuristics-based screening process. The 163 candidate peptides were then simulated according to computer simulations and tested for cell membrane binding propensity, which identified 20 lead candidate peptides that showed high and consistent cell membrane binding activity in the computer simulations. brought The 20 lead candidate peptides were then synthesized and tested for antimicrobial activity and toxicity using wet laboratory experiments. Of these 20 lead peptides, two final lead AI peptide designed peptides were identified. These two final lead AI-designed peptides were experimentally confirmed to have strong broad-spectrum antimicrobial activity and low in-vitro and in-vivo toxicity. Both of these novel AMPs were absent in the supervised training data used to design the initial candidate peptides. These experiments demonstrate that the disclosed three-stage screening pipeline (e.g., heuristics screening, simulation screening, and wet laboratory screening) against AI-generated AMP sequences yields 1 out of 10 successes in the final stage. Indicates to get the rate.

As used herein, the term "AI-designed molecule" refers to one or more of machine learning (ML) or artificial intelligence (AI) or both designed, generated or otherwise Browse developed molecules. The disclosed AI-designed molecules include biological molecules (e.g., natural and recombinant peptides, proteins, biopolymers, nucleic acids, polysaccharides, antibodies, hormones, etc.), synthetic molecules, biopharmaceuticals (or "biopharmaceuticals"), and combinations thereof. The disclosed AI-designed molecules can comprise organic compounds, inorganic compounds, compositions, organometallic compounds, or combinations thereof.

The term "peptide" as used herein refers to a polymer of amino acid residues typically ranging from 2 to about 50 residues. In certain embodiments, the AI-designed peptides disclosed herein range in length from about 2-25 residues. In some embodiments, amino acid residues comprising peptides include "L-form" amino acid residues, however, it is recognized that in various embodiments, "D" amino acids can be included in peptides. be done. Peptides also include amino acid polymers wherein one or more amino acid residues are naturally occurring amino acid polymers, as well as artificial chemical analogues corresponding to the corresponding naturally occurring amino acids. is.

As used herein, the term “synthetic” peptide or synthetic AMP is used to refer to peptides that are chemically synthesized as opposed to being host-derived.
As used herein, the term "residue" refers to natural, synthetic, or modified amino acids. Various amino acid analogs include, but are not limited to, 2-aminoadipic acid, 3-aminoadipic acid, β-alanine (β-aminopropionic acid), 2-aminobutyric acid, 4-aminobutyric acid, piperidic acid, 6 -aminocaproic acid, 2-aminoheptanoic acid, 2-aminoisobutyric acid, 3-aminoisobutyric acid, 2-aminopimelic acid, 2,4-diaminobutyric acid, desmosine, 2,2'-diaminopimelic acid, 2,3-diaminopropionic acid, n -ethylglycine, n-ethylasparagine, hydroxylysine, allohydroxylysine, 3-hydroxyproline, 4-hydroxyproline, isodesmosine, alloisoleucine, n-methylglycine, sarcosine, n-methylisoleucine, 6-n-methyllysine, n - Including methylvaline, norvaline, norleucine, ornithine, etc. These modified amino acids are exemplary and not intended to be limiting.

"Conventional" and "natural" as applied herein to peptides refer to the naturally occurring amino acids: Ala, Cys, Asp, Glu, Phe, Gly, His, Ile, Lys, Leu, Met, Asn, Reference is made to peptides composed only of Pro, Gln, Arg, Ser, Thr, Val, Trp, and Tyr. In various embodiments, the disclosed AI-designed peptides comprise only natural amino acid residues. In some embodiments, the disclosed AI-designed molecules can substitute one or more synthetic or modified amino acids for the corresponding naturally occurring amino acids. A compound of the present invention is a "natural peptide" if it induces a biological activity (e.g., antimicrobial activity) associated with the biological activity or specificity of the naturally occurring peptide, or both. handle". The induced activity may be the same, higher or lower than that of the native peptide. Generally, such peptides will have essentially corresponding monomer sequences, where if the N-substituted glycine derivative resembles the parent amino acid in hydrophilicity, lipophilicity, polarity, etc. , the natural amino acid is replaced by an N-substituted glycine derivative.

Also contemplated in certain embodiments are AMPs that have at least 80%, at least 85% or 90%, and more preferably at least 95% or 98% sequence identity with any of the sequences described herein. be. The terms "identical" or percent "identity" refer to amino acid residues when compared and aligned for maximum correspondence as determined using one of the following sequence comparison algorithms or visual inspection: Reference is made to two or more sequences that are identical or have a specified percentage of identical amino acid residues. For peptides disclosed herein, sequence identity is determined over the entire length of the peptide. For sequence comparison, typically one sequence acts as a reference sequence against which test sequences are compared. When using a sequence comparison algorithm, test and reference sequences are entered into a computer, subsequence coordinates are specified if necessary, and sequence algorithm program parameters are specified. The sequence comparison algorithm then calculates percent sequence identities for the test sequence(s) relative to the reference sequence based on specified program parameters. Optimal alignment of sequences for comparison can be performed using a basic local alignment search tool (BLAST) or the like.

The term "specificity" when used in reference to the antimicrobial activity of a peptide indicates whether the peptide preferentially inhibits growth or mitotic proliferation or both relative to other related species, or a combination of these. In certain embodiments, preferred inhibition or eradication is at least greater than 10% (e.g., LD50 below 10%), preferably 20%, 30%, 40%, or 50%, or more preferably is at least 2-fold, at least 5-fold, or at least 10-fold for the target species.

As used herein, "treatment" or "treatment" of a disease includes preventing the disease, slowing the onset or rate of progression of the disease, reducing the risk of progression of the disease, Refers to preventing or slowing the progression of symptoms, relieving or ending symptoms associated with the disease, producing complete or partial recovery from the disease, or some combination thereof .

The term "high," as used herein with respect to antimicrobial activity or efficacy, or both, means that the level of antimicrobial activity of an antimicrobial agent (e.g., AMP, etc.) is effective against a particular bacterial organism. Used to indicate that microbial activity or efficacy is greater than a defined minimum threshold. In various embodiments, the minimum threshold can be based on its MIC, its LD50 concentration/or HC50 concentration, wherein decreasing concentration increases antimicrobial activity or efficacy or both. For example, in some embodiments, the antimicrobial agent has a MIC of less than 250 micrograms per milliliter (μg/mL), more preferably less than 150 μg/mL), more preferably less than 100 μg, more preferably , 50 μg, and more preferably less than 30 μg/mL can be considered to have high antimicrobial activity or efficacy or both.

The term "low toxicity" as used herein refers to any level of toxicity of a drug (eg, one or more AMPs or another active agent) that is below a defined acceptable threshold of toxicity. In various embodiments, the defined threshold can be based on the drug's MIC for its LD50 or HC50 concentration or both. In some implementations, a drug (e.g., AMP or a composition comprising one or more AMPs) has low toxicity if its MIC is lower than its LD50 or HC50 concentration or both. can be thought of as In other implementations, a drug can be considered of low toxicity if its MIC is 60% or less of the LD50 or HC50 concentration or both. In other implementations, a drug can be considered of low toxicity if its MIC is 50% or less of the LD50 or HC50 concentration or both. In other implementations, a drug can be considered of low toxicity if its MIC is 30% or less of the LD50 or HC50 concentration or both. In other implementations, a drug can be considered of low toxicity if its MIC is 25% or less of the LD50 or HC50 concentration or both.

Various embodiments of the disclosed subject matter are illustrated for evaluating AI-designed molecules, and more specifically AI-designed AMPs, that are (or are intended to be) novel pharmaceutical agents. However, the disclosed filtering technique for AI-designed molecules can be applied to various target classes (antivirals, anti-neoplastics, therapeutics, anti-neoplastics), in addition to novel molecules designed for non-medical use. It should be recognized that it can be used to evaluate a variety of pharmaceutical agents that have specific properties for drugs, etc.). The terms "pharmaceutical", "pharmaceutical", "drug", "medicine" and "bioactive molecule" are used herein to refer to diagnostic, curative, therapeutic, or used interchangeably to refer to a substance that is used (or designed to be used) for disease prevention.

One or more embodiments are now described with reference to the drawings, wherein like numerals are used to refer to like elements throughout. In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a more thorough understanding of one or more embodiments. It is evident, however, that in various instances one or more embodiments may be practiced without these specific details. It should be noted that the drawings of the present application are provided for illustrative purposes only and, as such, the drawings are not drawn to scale.

FIG. 1 illustrates a high-level flow diagram of an example pipeline 100 for filtering candidate AI design molecules in accordance with one or more embodiments. Pipeline 100 prepares an initial set of candidate AI design molecules 102 (also referred to herein as "candidate molecules" or simply as "candidates") into one or more available candidates 114 Use a three-phase screening regime to filter to The three phases include a heuristics-based screening phase 104 , a computer simulation screening phase 108 and a wet laboratory screening phase 112 . According to the pipeline 100, a heuristics-based screening phase 104 selects the first candidate from the initial set 102 based on one or more predefined target features using one or more classifiers. It is used to select the subset 106 of 1. A computer simulation screening phase 108 then uses physically driven computer simulations to evaluate the relevant molecular dynamics of each candidate in the first subset, It is used to select a second subset 110 of lead candidate AI-designed molecules from the subset 106 . For example, a computer simulation may be performed on each candidate (included within the first subset 106) and one or more molecular/biological targets of the AI-designed molecule (e.g., one or more of pathogens). (cellular components) can be simulated. A second subset 110 is then selected based on whether and/or to what extent the candidates exhibit one or more target behavioral characteristics in the computer simulation.

A wet laboratory screening phase 112 is then used to screen each candidate contained in the second subset 110 (also referred to herein as a lead candidate) to remove any available identify possible candidates 114 . In various embodiments, the wet laboratory screening phase 112 includes synthesizing lead candidates and performing appropriate in-vitro or in-vivo testing or both, wherein the lead candidates undergo heuristics-based As indicated based on the screening phase 104 and the computer simulation screening phase 108, the availability against one or more pathogens or other molecular targets is confirmed. For example, in one or more embodiments in which the AI-designed molecules include molecules designed to be used as antimicrobial agents (e.g., AMPs), the wet laboratory screening phase 112 includes ( (including but not limited to) testing the lead candidate against one or more types of Gram-positive bacteria or Gram-negative bacteria or both or another type of pathogen; It can include testing in vitro or in-vivo. Additional details regarding the AI Design Molecule filtering pipeline (eg, pipeline 100) are further described with reference to FIGS.

FIG. 2 illustrates a block diagram of an example, non-limiting system 200 for filtering AI-designed molecules for wet laboratory testing described herein according to one or more embodiments. Embodiments of the systems described herein may include one or more machine-executable components embodied within one or more machines (e.g., one or more one or more computer-readable media associated with the machine). Such components, when executed by one or more machines (e.g., processors, computers, computing devices, virtual machines, etc.), cause one or more machines to perform the operations described. be able to.

For example, in the illustrated embodiment, the system 200 includes a heuristics-based screening component 202 and a simulation-based screening component 202, each of which is a machine, or corresponding to a machine, or a computer-executable component. Includes component 204 . System 200 may further include or be operatively coupled with at least one memory 210 and at least one processor 208 . In various embodiments, the at least one memory 210 stores executable instructions (e.g., heuristics) that, when executed by the at least one processor 208, facilitate performing operations defined by the executable instructions. based screening component 202, simulation based screening component 204, and additional components described below). System 200 may further include a device bus 206 that communicatively couples the various components of system 200 . Examples of the processor 208 and memory 210 described above, along with other suitable computer or computing-based elements, can be found with reference to FIG. 12 for the processing unit 1216 and system memory 1214, and It can be used in connection with implementing one or more of the systems or components shown and described in connection with FIG. 1 or other figures described herein.

In some embodiments, system 200 includes a processor or any type of component that is enabled or operable to communicate with or capable of wired or wireless networks or both. , machine, device, function, apparatus or instrument, or any combination thereof. All such embodiments are envisioned. For example, system 200 may include a server device, computing device, general purpose computer, special purpose computer, tablet computing device, handheld device, server class computing machine, or database or combination thereof; laptop computers, notebook computers, desktop computers, mobile phones, smart phones, consumer electronics or appliances or both, industrial or commercial devices or both, digital assistants, multimedia internet capable telephones, It may be located or operated or performed by or in combination with a multimedia player or another type of device.

The disclosed embodiments of the subject matter illustrated in the various figures disclosed herein are for purposes of illustration only, and thus the architecture of such embodiments may be applied to the systems, devices, and systems illustrated herein. or components or combinations thereof. In some embodiments, one or more components of system 200 are executed separately or in parallel by different computing devices (e.g., virtual machines) according to a distributed computing system architecture. can be done. System 200 may also include operating environment 1200 and various additional computer or computing-based elements, or combinations thereof, described herein with reference to FIG. In some embodiments, such computers or computer-based elements are the systems, devices, components, components shown and described in connection with FIG. 1 or other figures disclosed herein. or computer-implemented operations, or in connection with implementing one or more of these operations.

In some embodiments, system 200 connects to a data cable (e.g., High Definition Multimedia Interface (HDMI)), Recommended Standard (RS) 232, Ethernet cable, etc. ) to one or more external systems, data sources, or devices or combinations thereof (eg, communicatively, electrically, or optically).
In other embodiments, system 200 can be coupled (e.g., communicatively, electrically, or optically).

In many embodiments, such networks include, but are not limited to, cellular networks, wide area networks (WAN) (eg, the Internet), or local area networks (LAN). and wireless networks. For example, the heuristics-based screening component 202 or the simulation-based screening component 204 or both can be used for wireless fidelity (Wi-Fi), global system for mobile communications (GSM), universal mobile Telecommunications System (UMTS), Worldwide Interoperability for Microwave Access (WiMAX), Enhanced General Packet Radio Service (Enhanced GPRS), 3rd Generation Partnership Project (3GPP) Trademark)), Long Term Evolution (LTE), 3rd Generation Partnership Evolution 2 (3GPP®2), Ultra Mobile Broadband (UMB), High Speed Packet Access (HSPA), Zigbee ( trademark) and other 802. XX wireless technology or legacy telecommunications technology, BLUETOOTH®, Session Initiation Protocol (SIP), ZIGBEE®, RF4CE protocol, WirelessHART protocol, 6LoWPAN (IPv6 over low power wireless - Area Networks), Z-Wave, ANT, Ultra Wideband (UWB) standard protocols, or other proprietary or non-proprietary communication protocols or combinations thereof. Virtually can be used to communicate with one or more external systems, sources, or devices or combinations thereof, such as computing devices and vice versa. In such embodiments, system 200 may thus be hardware (eg, central processing unit (CPU), transceiver, decoder), software (eg, set of threads, set of processes, running software) or system 200 may include a combination of hardware and software that facilitates communication of information between the device and external systems, sources, or devices, or combinations thereof.

The system 200 facilitates filtering a large data set of AI-designed molecules into a significantly smaller data set of more targeted and promising candidates (i.e., second-order candidates for AI-designed molecules). subset), these likely provide target activities/functions for more complete confirmatory experiments, such as wet laboratory experiments, clinical trials of new pharmaceuticals, and the like. To facilitate this outcome, system 200 can include heuristics-based screening component 202 and simulation-based screening component 204 .

Referring again to FIG. 1 and looking at FIG. 2, heuristics-based screening component 202 is configured to perform heuristics-based screening phase 104 of pipeline 100 to identify the first candidate AI design molecule. 1, the simulation-based screening component 204 is configured to perform a computer simulation screening phase 108 of the pipeline 100 to generate a subset 106 of candidate AI design molecules. Allowing to generate a second subset 110 . As shown in FIG. 1, the output of system 200 includes a second subset 110 of candidate AI design molecules, which are recommended available for further testing (e.g., wet laboratory testing). corresponds to a reduced set of possible candidates.

In this regard, the system 200 can receive (or otherwise access) an initial set of candidate AI design molecules 102 for screening/filtering. The initial set 102 of candidate AI design molecules can include any number of candidate molecules (eg, including hundreds to thousands to hundreds of thousands or more). The types of AI-designed molecules included in the initial set or their target biological or chemical activities or combinations thereof can vary. In some embodiments, the initial set of candidate AI-designed molecules 102 are designed to provide a diagnostic, therapeutic, curative, or specific biological response in relation to a particular disease or combination thereof. Including pharmaceuticals. For example, the initial set of candidates 102 can include AI-designed molecules designed for function as active antimicrobial agents, antiviral agents, anticancer agents, and the like. In another more specific embodiment, the system 200 can be configured to screen AI-designed peptides specifically designed to function as broad spectrum antimicrobial peptides. According to this embodiment, the candidate initial set 102 of candidate AI design molecules can comprise a collection of such peptides.

In some embodiments, the initial set of candidates 102 can vary with respect to their molecular sequence, or their chemical structures still share common design factors or share another common attribute. Can be any combination of these. For example, in some implementations, the initial set of candidates 102 can include molecules generated/designed using one or more of the same ML/AI design models. In another embodiment, the initial set of candidates is designed to provide the same or similar target biological/chemical activity or function, or the same or similar biological/molecular target. or combinations thereof. Additionally or alternatively, the initial set of candidates 102 may include a collection of AI-designed molecules that vary with respect to one or more of these common factors, such as randomly sampled AI-designed molecules. .

Regardless of the distribution of AI-designed molecules included in the initial set 102, the heuristics-based screening component 202 and the simulation-based screening component 204 may perform target biological activity/function or target chemical activity/function. can be configured to screen candidates based on For example, in implementations where the target biological activity/function provides broad-spectrum antimicrobial activity (e.g., activity against both Gram-positive and Gram-negative characteristics), a heuristics-based screening component 202 and simulation-based screening component 204 are configured to screen candidates to a small subset of the most available candidates expected to provide broad-spectrum antimicrobial activity (e.g., A second subset 110) of candidate AI design molecules can be selected. Additional details of heuristics-based screening component 202 are described with reference to FIGS. 3A and 3B and FIG. Additional details of simulation-based screening component 204 are described with reference to FIGS. 5A-9.

3A and 3B show block diagrams of example heuristics-based screening components in accordance with one or more embodiments. Respective descriptions of similar elements utilized in each embodiment are omitted for the sake of brevity.

According to the embodiment shown in FIG. 3A, heuristics-based screening component 202 includes classifier application component 302 , first subset selection component 304 , and one or more classifiers 306 . In various embodiments, the apply classifier component 302 applies one or more classifiers to the initial set of candidate AI design molecules 102 to determine their respective molecular sequences (e.g., protein sequences, Each (or in some implementations, one or more) of the initial candidate molecules are defined targets based on analysis of their gene/nucleotide sequence, polymer sequence, etc.) or their chemical structure or both. It can be configured to determine or infer whether it has one or more of the features (ie features of interest). In this regard, the heuristics-based screening phase is based on the analysis and classification of candidate molecules at the sequence level or chemical structure level or both.

One or more defined target features can be pre-selected and one or more of the target AI-designed molecules that the disclosed filtering techniques will be used to identify. It reflects the above desired features. The one or more characteristics may include implicit characteristics known to correlate with explicit characteristics (e.g., having a secondary peptide structure associated with antimicrobial activity), as well as explicit characteristics (e.g., , exhibit antimicrobial activity, exhibit broad spectrum susceptibility). One or more of the target characteristics may thus vary based on the particular application of pipeline 100 or system 200 or both.

For example, in some embodiments, pipeline 100 or system 200, or both, are used to screen candidate AI-designed peptides, most likely to provide effective, broad-spectrum antimicrobial agents. Allows identification and selection of a small subset of candidate AI design peptides. In these embodiments, one or more of the defined characteristics include (but are not limited to) antimicrobial functionality, broad spectrum efficacy, low or no toxicity, efficacy, and defined structure. The presence (eg, secondary structures such as helical structures, pleated sheet structures, coil structures, etc.) can be included. One or more of the classifiers 306 therefore determine whether each of the initial candidate peptides has antimicrobial function (or not), broad spectrum efficacy (or not), low or no toxicity, It can be configured to predict being (or not), having defined secondary structure (or not), or having high potency (or not).

In some embodiments, one or more of the classifiers 306 classify in a defined target feature and/or a molecular sequence (e.g., protein sequence) or chemical structure of a known molecule with the target feature. pre-trained to classify each candidate as either having one or more defined target features or not based on learned correlations between reflected patterns; It can contain one or more binary classification models. In other implementations, one or more of the classifiers 306 determine the probability that the candidate molecule has each target feature (e.g., the probability that it has target feature 1, the probability that it has target feature 2, the probability that it has target feature 3, etc.). can be configured to predict the probability of having In some implementations, each classifier of one or more classifiers 306 can be trained to classify a single target feature. For example, for the AMP implementation described above, one or more of the classifiers 306 may classify each of the four target characteristics (e.g., antimicrobial function, broad spectrum efficacy, low or no toxicity, and presence of defined structures). can include up to four separate classifiers for one of

Various types of classification models/algorithms can be used for one or more classifiers 306 . In some embodiments, the one or more classifiers 306 are one or more deep learning neural networks, such as long short-term memory (LSTM) neural network-based classifiers. can include The heuristics-based screening component 202 can also facilitate classification of one or more target features of the initial candidate molecules using automated classification systems and/or processes. For example, the heuristics-based screening component learns about the initial set of candidate AI design molecules 102 using probabilistic or statistical-based analysis (e.g., factoring into analytical utility and cost) or both; or generate an estimate, or both. The heuristics-based screening component 202 can learn about the initial set of candidates 102 and/or generate estimates using, for example, a support vector machine (SVM) classifier.

Additionally or alternatively, one or more of the classifiers 306 may use classification techniques associated with Bayesian networks, decision trees, or probabilistic classification models or combinations thereof. One or more classifiers 306 may also be explicitly trained (e.g., genetically trained classifiers) in addition to implicitly trained (e.g., via receipt of extrinsic information). data) classifier. For example, for SVM's, the SVM' can be configured via a learning or training phase within the classifier constructor and feature selection module. In some implementations, the one or more classifiers 306 also classify the input attribute vector x = (x1, x2, x3, x4, xn) so that the input vector belongs to the classification, i.e. f(x ) = reliability (class). For these implementations, the apply classifier component 302 can determine a measure of confidence in predicting whether a candidate has each of the target features evaluated.

A first subset selection component 304 can be configured to select a first subset 106 of candidate AI design molecules from the initial set 102 based on the classification results and defined selection criteria. Selection criteria can be predefined and adjusted, such as by a system administrator. For example, in some implementations, the selection criteria select only those candidates for which the first subset selection component 304 determines that they have (or are classified as having) all of the defined target characteristics. can be requested. In another embodiment, the selection criteria select those candidates for which the first subset selection component 304 determines that they have (or are classified as having) one or more of the defined target characteristics. can be requested. In another embodiment, the selection criteria determine that the first subset selection component 304 has (or is classified as having) a particular combination of target features with one or more defined target features. can be requested to select those candidates that have been presented. In another embodiment, in implementations in which one or more classifiers 306 determine values representing the probabilities that candidate molecules have respective probabilities, the selection criteria are probabilities over all features or collective It can include defined thresholds for scores representing probabilities or both.

It should be appreciated that the selection criteria can be set appropriately for a particular application (e.g., the number of defined features required, the combination of features required, the level of representation of features, etc.). value, a value that indicates the degree of confidence in the classification estimate, etc.).

FIG. 3B presents another embodiment of heuristics-based screening component 202 . In the embodiment shown in FIG. 3B, the heuristics-based screening component 202 further includes a classifier training component 308 to facilitate training and developing one or more classifiers 306. . For these embodiments, the classifier training component 308 is received or otherwise available using one or more unsupervised, supervised, or semi-supervised machine learning techniques or combinations thereof. Based on such training data 310, one or more classifiers 306 can be trained and developed. For example, the training data 310 includes a sequence of positive classifications (e.g., with one or more specific target features) and negative classifications whose classifications for one or more of the target features are known. A plurality of molecular sequences (eg, protein sequences) can be included, including (eg, without one or more specific target features). Using the set of positive and negative sequences for each target feature, classifier training component 308 can train a separate classifier for each target feature.

FIG. 4 provides a table 400 showing example heuristic classification results for candidate antimicrobial peptides according to one or more embodiments. In particular, table 400 is generated or determined by classifier application component 302 based on application of five different classifiers to a plurality of candidate AMP sequences based on their respective peptide sequences shown in the first column. Or we present example heuristic classification data that can do both. Each of the five different classifiers is identified by the symbol "clfX_feature", where "clf" is a prefix and "X" is the specific training dataset used to train the classifier. show.

The first classifier, clfX.amp (where "amp" stands for antimicrobial peptide), measures the probability (0.0-1 .0) was determined. A second classifier clfX.tox (where "tox" stands for "toxicity") determined the probability (0.0-1.0) that a peptide sequence was toxic. A third classifier, clfX.Validity, determined the probability (0.0-1.0) that the peptide sequence was valid. A fourth classifier, clfX.borad (where "broad" stands for "broad spectrum"), measures the probability (0.0-1.0) that a peptide sequence is broad-spectrum antimicrobial. judged. A fifth classifier clfX.struct (where "struct" stands for "structure") determined the probability (0.0 to 1.0) that a peptide sequence has secondary structure.

5A and 5B show block diagrams of example simulation-based screening components in accordance with one or more embodiments. Respective descriptions of similar elements utilized in each embodiment are omitted for the sake of brevity.

The simulation-based screening component 204 uses a high-throughput, computationally efficient, physically-implied filtering process using physics-based molecular computer simulations to identify the first AI-designed molecules. is provided for further refinement into a second, smaller subset 110 of candidate AI design molecules to recommend for wet laboratory testing. These computer simulations compare each candidate in the first subset 106 with one or more known or potential molecular and/or biological targets (e.g., one of a pathogen or more cellular components) to determine whether or not the simulated candidates exhibit one or more desired interaction properties or to what extent or both. . In this regard, one or more desired interactions (or desired behavioral characteristics) are targeted biological/molecular activities, functions or responses (e.g., antimicrobial activity, antiviral activity, specific can include one or more predefined or learned or both interaction behaviors/characteristics that are correlated with achieving the therapeutic activity of the drug. For example, in implementations involving a target biology/molecular activity/response that would be an effective antimicrobial agent, one or more desired interaction/behavioral properties would eradicate bacteria or It can include one or more molecular interaction behavioral characteristics that correlate with growth inhibition or both.

Referring to FIG. 5A, to facilitate this conclusion, the simulation-based screening component 204 includes a simulation execution component 502, a simulation evaluation component 504, one or more simulation programs 506, and a second It includes a subset selection component 508 .

One or more simulation programs 506 can include one or more high-throughput computer simulation programs capable of simulating physics-based molecular interactions. In particular, the one or more simulation programs 506 simulate the molecular interactions between the AI-designed molecule and one or more biological/molecular targets based on their modeled molecular or Molecular simulation tools can be provided that can simulate based on biological or both structures. For example, these simulation tools can include coarse-grained molecular dynamics (CGMD) simulation tools, and the like. For example, in some implementations, one or more simulation programs 506 receive and/or generate molecular models for each candidate molecule included in the first subset 106. can contain. In some implementations, the molecular model can include an all-atom model. One or more simulation programs 506 are further modeled as force fields for biological/molecular target(s) (e.g., one or more cellular components of a pathogen) ( It can receive or generate a molecular model (eg, a coarse-grained force field, etc.) or both. The one or more simulation programs 506 may further apply a coarse-grained system for combinations of molecular candidates and biological/molecular target(s) (e.g., one or more cellular components of a pathogen). Along with generating a representation, this coarse-grained system representation can be used to simulate the molecular dynamics of the interaction between each candidate and the biological/molecular target(s).

Simulation execution component 502 can be configured to run/run one or more simulations for each candidate in first subset 106 . In this regard, the simulation execution component 502 can run a CGMD for each (or in some implementations one or more) candidate AI design molecules contained in the first subset 106, where: Each simulation is based on each modeled molecular structure, modeled using one or more force field models, and one or more defined Simulate molecular dynamics between biological/molecular targets.

A simulation evaluation component 504 evaluates each simulation such that each simulated candidate AI design molecule (i.e., each candidate molecule included in the first subset 106) has one or more target molecular interactions. / behavioral characteristics, and/or to what extent. For example, in some implementations, the molecular simulation program used is configured to identify and track the occurrence of one or more target molecule interaction/behavioral properties over each simulation trajectory. can be For these embodiments, the simulation program can generate result data indicating whether, for each simulation, one or more target molecule interactions/behavioral properties occurred, how often they occurred, etc. .
The simulation evaluation component 504 further uses the result data generated for each simulation to evaluate each simulated candidate AI design molecule (i.e., each candidate molecule included in the first subset 106) in one or more It can be determined whether or not, or to what extent, or both, the above target molecule interaction/behavioral properties are exhibited. In other embodiments, the simulation is manually observed and evaluated to see if each simulated candidate AI design molecule exhibits one or more target molecule interaction/behavioral properties. , or the degree thereof, or both, For these embodiments, such result data may be received as user-generated feedback.

A second subset selection component 508 further selects a second subset based on whether and/or to what extent each simulated candidate molecule exhibits one or more target molecule interaction/behavior properties. Select one or more simulated candidate molecules for inclusion in the subset 110 of . For example, in some implementations, the second subset selection component 508 is configured to select any simulation candidates determined to exhibit molecular interaction/behavioral properties of one or more targets. can In other implementations, the second subset selection component 508 is determined to exhibit one or more target molecule interaction/behavior characteristics that are consistent, or have sufficient propensity, or both. It can be configured to select one or more simulated candidates (eg, for a defined threshold to measure consistent or sufficient trend). In another example implementation, the second subset selection component 508 selects the "best" one or more target molecule interaction/behavior properties as measured using the defined evaluation scheme. can be configured to select one or more simulated candidates determined to be shown in . In this regard, evaluation schemes and selection criteria can vary based on the type of molecular interaction/behavioral properties evaluated or the manner in which they can be measured.

In one or more exemplary embodiments in which the candidate AI design molecules are candidate AMPs, to screen candidate peptides for potential antimicrobial properties, simulation execution component 502 first A computer simulation (eg, a CGMD simulation, etc.) of the interaction between each of the candidate peptides contained in the subset 106 of the pathogen and a model lipid bilayer or another cellular component of the pathogen can be run. A lipid bilayer can be composed of a mixture of lipids. For example, a candidate peptide can be modeled with a suitable all-atom representation (eg, alpha helix or random coil) of the peptide of that given protein sequence. A model lipid bilayer can be further modeled using a force field model, such as a coarse-grained force field model. The modeled peptide structure can be further converted to a coarse-grained representation and combined with the cell membrane model to generate a coarse-grained peptide-cell membrane system for simulation.

For example, FIG. 6 provides a snapshot of a coarse-grained molecular dynamics simulation of AMPs according to one or more embodiments. In this simulation, the modeled peptides were bound to a modeled lipid bilayer, which in this example simulated phosphatidylcholine (POPC) and palmitoyl-oleoyl PG (POPG) 3: 1. FIG. 6 illustrates a CGMD simulation using modeled peptides and modeled cell membranes. According to these simulations, each candidate peptide interacts with the cell membrane for 1.0 microseconds (μs). The physical kinetics of this interaction is then evaluated to determine whether the interaction indicates that the peptide provides antimicrobial activity.

In one or more embodiments, the target interaction/behavior used to assess antimicrobial propensity based on the computer simulations described above is the contact/touch point between the peptide and the cell membrane. number, and the stability of their contacts. In this regard, as explained in more detail with reference to FIG. 5B, the antimicrobial propensity was found to be strongly correlated with the number of contacts and the stability of the contacts, where the number of contacts increased. , and the greater the stability of these contacts, the greater the potential for antimicrobial propensity. Contacts can include contacts between the positive residues of the peptide and the cell membrane. In one or more implementations, the number of contacts between the positive residue and the lipid cell membrane is defined as the number of atoms contained in the lipid that are less than 7.5 Å from the positive residue of the peptide. . Contact stability can be measured as a function of the difference in the number of contacts, where less difference is a higher stability and thus a higher indicator of potent antimicrobial activity.

FIG. 7 provides a table 700 presenting example simulation results for candidate AMPs in accordance with one or more embodiments. Table 700 provides computer simulation results for a plurality of example candidate peptide sequences each identified in the first column. The peptide lengths, their respective secondary structures and the number of positive residues for each sequence are included in the second, third and fourth columns, respectively. The fifth column provides the standard deviation (std) of the number of contacts, which corresponds to the difference in the number of contacts. The sixth column provides the average number of contacts. The seventh column provides the binding time in nanoseconds (ns). Binding time is the time interval for the peptides to form contacts after the start of the simulation. In the illustrated embodiment, all example peptides made their contacts in less than 500 (ns), which is preferred and can also be used as a filtering criterion.

Referring again to FIG. 5A and looking at FIG. 7, in facilitating the AMP candidate screening embodiment, the simulation evaluation component 504 calculates the number of contacts between the lipids and the positive residues of each candidate peptide and the contact It is possible to determine and/or receive simulation results (provided in table 700) that identify differences in the number of . In some implementations, the simulation results can also include binding times, which can also be used as filtering criteria as described above. The second subset selection component 508 further includes one or more subgroups exhibiting a propensity for consistent cell membrane interactions as determined based on number of contacts, difference value, or binding time or a combination thereof. Candidate peptides can be selected. For example, in one or more embodiments, the second subset selection component 508 can use the defined difference acceptable criteria and their difference value, number of contacts, or combination Only those candidate peptides meeting defined acceptance criteria for time or a combination thereof can be selected. In some implementations, defined acceptability criteria are the number of differences (i.e., standard deviation) less than 2.0 beads, the number of contacts greater than or equal to 5.0 (average over the period of the simulation), and It may be required that their coupling time during a simulation time of 1.0 μs length be less than 500 ns (eg, so that contact differences are calculated over at least half of the total simulation time).

Referring now to FIG. 5B, presented is another example of simulation-based screening component 204 in accordance with one or more additional embodiments. Each description of each similar element or process or combination thereof utilized in each embodiment is omitted for purposes of brevity.

In the above-described embodiment directed to simulation-based screening of candidate AMPs, the examples and target molecule interaction features/behaviors that we evaluated and used to select a second subset of candidate AI design molecules are: , the number of contacts/touch points between the peptide and the cell membrane, and the stability of these contacts (measured in the difference in number of contacts). These target features are identical as applied to known peptide sequences known to have antimicrobial properties and peptide sequences known to lack antimicrobial activity. The discovery was made by running test simulations using molecular modeling simulations of molecular simulations, due to the complete lack of standardized protocols for screening antimicrobial candidates using molecular simulations.

Based on the analysis of the results of running tests on both positive and negative antimicrobial peptides, the specific target features described above were first identified. In this regard, test simulation behavior showed that differences in the number of contacts between positive residues and cell membrane lipids were predictive of antimicrobial activity.

In particular, FIG. 8 presents a confusion matrix 600 of an example simulation-based classifier that uses peptide-cell membrane contact differences as a feature to detect available AMP sequences. The confusion matrix 600 shows that by using the contact difference features derived from the simulation alone described above, we can predict antimicrobial activity with 88% accuracy. In particular, contact differences discriminate between high efficacy with a sensitivity of 88% and a specificity of 63% and non-antimicrobial sequences. Physically this feature can be interpreted as measuring the robust binding of the sequence to model cell membranes.

In various embodiments, this test simulation process is performed or facilitated by or facilitated by simulation-based screening component 204 using simulation execution component 502 and feature selection component 512 . both are done. This test simulation process can also be used to determine target characteristics for the process of simulation screening other types of AI-designed molecules for a variety of different target biological activities.

In this regard, in some embodiments, training a high-throughput computer simulation is performed to determine the target activity of an AI-designed molecule (e.g., a desired biological activity in implementations where the AI-designed molecule is a pharmaceutical). One or more behavioral tests that correlate with effectiveness in achieving a target activity are performed on test molecules known to be effective in achieving and optionally including test molecules that are not effective characteristics can be identified. These one or more behavioral features are evaluated (e.g., by the simulation evaluation component 504) and selected (e.g., by the simulation evaluation component 504) and selected ( for example, by the second subset selection component 508).

With these embodiments, the simulation execution component 502 corresponds to an initial set of candidate AI design molecules, or more specifically, their biological activity status is known (e.g., antimicrobial activity/ Inactive state) Test molecules 510 corresponding to a first subset of candidate AI design molecules can be received (or otherwise accessed). In this regard, test molecules 510 can include both molecules known to provide target biological activity and molecules known not to provide target biological activity. Simulation execution component 502 is further configured to apply the same computer simulation (eg, provided by simulation program 506) to test molecules 510 that would be used for first subset 106; can be done. The simulations for the test molecules are further evaluated to correlate one or more of the target biological activities (e.g., antimicrobial activity, antiviral activity, etc.) expected to be provided by the AI-designed molecule to be evaluated. Identify the above target features/characteristics. For example, for the AMR simulation embodiment described above, the features selected included differences in the number of contacts. Once identified, these features are then used to classify them based on target features (e.g., number of contacts between lipids and positive residues of peptides), as well as candidates for laboratory testing. Can be used to select the second subset 110 .

In the embodiment of FIG. 5B, simulation-based screening component 204 further includes feature selection component 512 to facilitate identification of these target features based on analysis of test simulations for positive and negative test molecules. can be In this regard, the feature selection component 512 uses one or more machine learning techniques to determine the target biological activity (e.g., antimicrobial activity) of the AI-designed molecule to be evaluated based on the test simulation data. , antiviral activity, etc.) can be identified. Machine learning techniques may include supervised machine learning techniques, semi-supervised machine learning techniques, unsupervised machine learning techniques, or combinations thereof. For example, machine learning techniques include expert systems, fuzzy logic, SVMs, hidden Markov models (HMMs), greedy search algorithms, rule-based systems, Bayesian models (e.g. Bayesian networks), neural networks, etc. non-linear training techniques, data fusion, utility-based analysis systems, systems using Bayesian models, etc., as well as the use of various classification techniques described herein.

FIG. 9 depicts a high-level flow diagram of an example, non-limiting, computer-implemented method 900 for filtering AI-designed molecules for laboratory testing in accordance with one or more embodiments. Each description of each similar element or process or combination thereof utilized in each embodiment is omitted for purposes of brevity.

At 902, a system (e.g., system 200) operatively coupled to a processor identifies one or more classifiers (e.g., heuristics-based screening component 202) by the system operatively coupled to the processor. ) is used to select a first subset of artificial intelligence (AI)-designed molecules from the set of AI-designed molecules as candidate drugs based on the classification of the AI-designed molecules. At 904, the system uses one or more computer simulations (eg, using simulation-based screening component 204) to identify drug candidates and one or more biological targets (eg, , one or more cellular components of the pathogen) to select a second subset of candidate drugs for wet laboratory testing.

FIG. 10 is a high-level flow diagram of an example, non-limiting, computer-implemented method 1200 for filtering candidate AI-designed antimicrobial molecules for laboratory testing in accordance with one or more embodiments. indicate. Each description of each similar element or process or combination thereof utilized in each embodiment is omitted for purposes of brevity.

At 1002, a system (such as system 200) operatively coupled to a processor determines whether the first AI-designed molecule is one of: AMP, broad-spectrum antimicrobial, non-toxic, or structured; Select a first subset of first artificial intelligence (AI)-designed molecules from the set of AI-designed molecules based on the determination of the AI-designed molecules to be greater (e.g., using heuristics-based screening component 202). can be done). For example, in one or more embodiments, heuristics-based screening component 202 uses one or more trained classifiers to classify each (or in some implementations) included in the initial set. whether the candidate AI design molecule is AMP, broad spectrum, toxic, or structured, or combinations thereof. Whether there is can be determined as described with reference to FIGS. 3A, 3B, and 4 . At 1004, the system selects a first subset for wet laboratory testing based on a second determination that the second AI-designed molecule has a defined level of propensity to interact with cellular components of the pathogen. can select a second subset of second AI designs from (eg, using simulation-based screening component 204). For example, in one or more embodiments, as described above with reference to FIGS. or another cellular component), using one or more computer simulations of molecular dynamics for each of the candidate peptides in the first subset, the propensity of their interaction as a function of contact differences. can be judged.

The screening techniques described herein have proven successful when applied to screen thousands of AI-designed AMPs to identify available candidates. In particular, the disclosed screening techniques were applied to an initial set of approximately 100,000 candidate peptides generated using an AI-based peptide design method referred to as Conditional Latent (attribute) Space Sampling, or CLaSS. rice field. The ClaSS design method uses attribute conditioning/control sampling from informative latent space learned using neural generative models to generate candidate AMPs.

An initial set of 100,000 candidate peptides was reduced to 163 candidate peptides using a heuristics-based screening process. To screen the initial 100,000 CLaSS-generated AMP sequences for experimental confirmation, four binary (yes/no) sequence-level neural • An independent set of network classifiers was used to predict antimicrobial function, broad spectrum efficacy (e.g., activity for both Gram-positive and Gram-negative features), presence of secondary structure, in addition to toxicity. A bidirectional LSTM-based classifier was trained on each of the four attributes on a labeled training dataset on known peptide sequences with a hidden layer size of 100 and a dropout of 0.3. rice field. Based on the distribution of scores (class probability/logit), thresholds were determined considering the 50th percentile (median) of scores. Screening criteria were used to select a first subset of candidates from the initial 100,000 available candidates, thus all four attributes were considered.
163 candidates passed this screen.

The 163 candidate peptides were then subjected to coarse-grained molecular dynamics (CGMD) simulations of peptide-cell membrane interactions to test for cell membrane binding propensity according to the simulation-based screening process described above. A simulation-based screen resulted in the identification of 20 lead candidate peptides that exhibited high and consistent cell membrane binding activity in computer simulations. These top 20 peptides have the following sequences (shown in 3-letter code, shown in brackets in 1-letter code):
Tyr Leu Arg Leu Ile Arg Tyr Met Ala Lys Met Ile (YLRLIRYMAKMI) (SEQ ID NO: 1),
Phe Pro Leu Thr Trp Leu Lys Trp Trp Lys Trp Lys Lys (FPLTWLKWWKWKK) (SEQ ID NO: 2),
His Ile Leu Arg Met Arg Ile Arg Gln Met Met Thr (HILRMRIRQMMT) (SEQ ID NO: 3),
Ile Leu Leu His Ala Ile Leu Gly Val Arg Lys Lys Leu (ILLHAILGVRKKL) (SEQ ID NO: 4),
Tyr Arg Ala Ala Met Leu Arg Arg Gln Tyr Met Met Thr (YRAAMLRRQYMMT) (SEQ ID NO: 5),
His Ile Arg Leu Met Arg Ile Arg Gln Met Met Thr (HIRLMRIRQMMT) (SEQ ID NO: 6),
His Ile Arg Ala Met Arg Ile Arg Ala Gln Met Met Thr (HIRAMRIRAQMMT) (SEQ ID NO: 7),
Lys Thr Leu Ala Gln Leu Ser Ala Gly Val Lys Arg Trp His (KTLAQLSAGVKRWH) (SEQ ID NO: 8),
His Ile Leu Arg Met Arg Ile Arg Gln Gly Met Met Thr (HILRMRIRQGMMT) (SEQ ID NO: 9),
His Arg Ala Ile Met Leu Arg Ile Arg Gln Met Met Thr (HRAIMLRIRQMMT) (SEQ ID NO: 10),
Glu Tyr Leu Ile Glu Val Arg Glu Ser Ala Lys Met Thr Gln (EYLIEVRESAKMTQ) (SEQ ID NO: 11),
Gly Leu Ile Thr Met Leu Lys Val Gly Leu Ala Lys Val Gln (GLITMLKVGLAKVQ) (SEQ ID NO: 12),
Tyr Gln Leu Leu Arg Ile Met Arg Ile Asn Ile Ala (YQLLRIMRINIA) (SEQ ID NO: 13),
Val Arg Trp Ile Glu Tyr Trp Arg Glu Lys Trp Arg Thr (VRWIEYWREKWRT) (SEQ ID NO: 14),
Leu Ile Gln Val Ala Pro Leu Gly Arg Leu Leu Lys Arg Arg (LIQVAPLGRLLKRR) (SEQ ID NO: 15),
Tyr Gln Leu Arg Leu Ile Met Lys Tyr Ala Ile (YQLRLIMKYAI) (SEQ ID NO: 16),
Tyr Gln Leu Arg Leu Ile Met Lys Tyr Ala Ile (HRALMRIRQCMT) (SEQ ID NO: 17),
Gly Trp Leu Pro Thr Glu Lys Trp Arg Lys Leu Cys (GWLPTEKWRKLC) (SEQ ID NO: 18),
Tyr Gln Leu Arg Leu Met Arg Ile Met Ser Arg Ile (YQLRLMRIMSRI) (SEQ ID NO: 19) and Leu Arg Pro Ala Phe Lys Val Ser Lys (LRPAFKVSK) (SEQ ID NO: 20),
and conventional modified variants thereof.

FIG. 11 provides a table 1100 displaying simulation results for the top 20 CLaSS-generated AMPs selected from 163 candidate peptides selected after a heuristics-based screening process. Table 1100 shows the averages and differences in the number of contacts between positive amino acids and beads on cell membranes, as extracted from CGMD simulations of peptide-cell membrane interactions (i.e., those relevant to microbial function). was found), presenting physically derived features of simulation-based screening. Criteria used to further filter the 163 candidates were a difference value (i.e., standard deviation) of less than 2.0 beads, a number of contacts greater than or equal to 5.0 (average over the period of the simulation), and a value of 1 Their coupling time was required to be less than 500 ns during a simulation time of .0 μs length. Based on a combination of CLaSS generation methods, ML heuristics-based screening process and molecular simulation results, these top 20 peptides exhibit potent antimicrobial activity or behavior and thus broad-spectrum antimicrobial activity. Antibiotics are promising. These top 20 peptides are further characterized as having low toxicity.

Top 20 lead candidate peptides were then synthesized and tested for antimicrobial activity and toxicity using wet laboratory experiments. Of these 20 lead peptides, two novel AMPs were identified with the highest antimicrobial activity. These two novel AMPs were experimentally confirmed to have potent broad-spectrum antimicrobial activity and low toxicity in-vitro and in-vivo. Both of the novel AMPs were absent from the supervised training data used to design the initial candidate CLaSS peptides. These experiments demonstrate that the disclosed three-stage screening pipeline for AI-generated AMP sequences (e.g., ML heuristics screening, simulation screening, and wet laboratory screening) yields indicates that it gives a success rate of 1.

Note that, for simplicity of explanation, in some environments a computer-implemented methodology is illustrated and described as a series of acts. The subject innovation is not limited by the acts shown or by the order of acts, or by any combination thereof; It is to be understood and appreciated that other operations not presented and described in can occur with. Moreover, not all illustrated acts may be required to implement a computer-implemented methodology in accordance with the disclosed subject matter. Additionally, those skilled in the art will understand and appreciate that a computer-implemented methodology may alternatively be represented as a state diagram or a series of interrelated states via events. Additionally, furthermore, the computer-implemented methodologies disclosed below and throughout this specification may be stored in an article of manufacture that facilitates sending and transferring such computer-implemented methodologies to computers. should be aware of what can be done. The term article of manufacture as used herein is intended to cover a computer program accessible from any computer-readable device or storage medium.

FIG. 12 can provide non-limiting context for various aspects of the disclosed subject matter and provides a general description of a suitable environment in which various aspects of the disclosed subject matter can be implemented. intended to FIG. 12 illustrates a block diagram of an example non-limiting operating environment that can facilitate one or more embodiments described herein. Each description of each similar element or process or combination thereof utilized in each embodiment is omitted for purposes of brevity.

Referring to FIG. 12, a suitable operating environment 1200 for implementing various aspects of the present disclosure can also include computer 1212 . Computer 1212 may also include processing unit 1216 , system memory 1214 , and system bus 1218 . A system bus 1218 couples system components including, but not limited to, system memory 1214 to processing unit 1216 . Processing unit 1216 can be any of a variety of available processors. Dual microprocessors and other multiprocessor architectures can also be utilized as processing unit 1216 . The system bus 1218 may be a memory bus or memory controller, peripheral or external bus, including but not limited to Industrial Standard Architecture (ISA), Micro Channel Architecture (MCA), Extended ISA (EISA), Intelligent Drive Electronics (IDE), VESA Local Bus (VLB), Peripheral Component Interconnect (PCI), Card Bus, Universal Serial Bus (USB), Advanced Graphics Port (AGP), Any of several types of bus structure(s), including any of the various available bus architectures including Firewire (IEEE 1394) and Small Computer Systems Interface (SCSI). can.

System memory 1214 may also include volatile memory 1220 and nonvolatile memory 1222 . The basic input/output system (BIOS), containing the basic routines to transfer information between elements within computer 1212 , such as during start-up, is stored in nonvolatile memory 1222 . Computer 1212 may also include removable/non-removable, volatile/non-volatile computer storage media. FIG. 12 shows disk storage 1224, for example. Disk storage 1224 may also include, but is not limited to, magnetic disk drives, floppy disk drives, tape drives, Jaz drives, Zip drives, LS-100 drives, flash memory cards, or memory sticks. can contain. Disk storage 1224 may also include storage media in isolation or in combination with other storage media. An interface, such as removable or non-removable interface 1226 is typically used to facilitate coupling disk storage 1224 to system bus 1218 . FIG. 12 also illustrates software that acts as an intermediary between users and the basic computer resources described in preferred operating environment 1200 . Such software may also include an operating system 1228, for example. An operating system 1228 can be stored in disk storage 1224 and operates to control and allocate the resources of computer 1212 .

System applications 1230 benefit from the management of resources by operating system 1228 , for example, through program modules 1232 and program data 1234 stored in either system memory 1214 or disk storage 1224 . It should be appreciated that the present disclosure can be implemented with different operating systems or combinations of operating systems. A user enters commands or information into computer 1212 through input device(s) 1236 . Input devices 1236 include, but are not limited to, pointing devices such as mice, trackballs, styluses, touchpads, keyboards, microphones, joysticks, game pads, satellite dishes, scanners, TV tuner cards, digital Including cameras, digital video cameras, web cameras, etc. These and other input devices connect to the processing unit 1216 through the system bus 1218 via interface port(s) 1238 . Interface port(s) 1238 include, for example, serial ports, parallel ports, game ports, and universal serial bus (USB). Output device(s) 1240 use some of the same type of port as input device(s) 1236 . Thus, for example, a USB port can be used to provide input to computer 1212 and output information from computer 1212 to output device 1240 . Output adapter 1242 is provided to illustrate that there are some output devices 1240 that require special adapters, such as monitors, speakers, and printers, among other output devices 1240 . Output adapters 1242 include, by way of example and not limitation, video and sound cards that provide a means of connection between output devices 1240 and system bus 1218 . It should be pointed out that other devices or systems of devices, or combinations thereof, provide both input and output functionality, such as remote computer(s) 1244 .

Computer 1212 can operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1244 . The remote computer(s) 1244 can be computers, servers, routers, network PCs, workstations, microprocessor-based appliances, peer devices or other common network nodes, etc., and are typically can also include many or all of the elements described for computer 1212 . For simplicity purposes, only memory storage device 1246 is shown for remote computer(s) 1244 . Remote computer(s) 1244 are connected locally to computer 1212 via network interface 1248 and then physically via communication connection 1250 . Network interface 1248 spans wired and/or wireless communication networks such as local-area networks (LAN), wide-area networks (WAN), cellular networks, and the like. LAN technologies include Fiber Distributed Data Interface (FDDI), Copper Distributed Data Interface (CDDI), Ethernet, Token Ring, and others. WAN technologies include, but are not limited to, point-to-point links, Integrated Services Digital Networks (ISDN) and variants thereon, packet-switched networks and Digital Subscriber Lines (DSL). include. Communication connection(s) 1250 refers to the hardware/software utilized to connect network interface 1248 to system bus 1218 . Communication connection 1250 is shown within computer 1212 for illustrative clarity, but it can also be external to computer 1212 . Hardware/software for connecting to network interface 1248 also includes, for illustrative purposes only, modems including landline grade modems, cable modems, and DSL modems, ISDN adapters and Ethernet cards. can include internal and external technologies such as

One or more embodiments can be systems, methods, or computer program products, or combinations thereof, in whatever level of technical detail possible. A computer program product may include a recording medium (or media) having computer-readable program instructions thereon to cause a processor to perform one or more aspects of the embodiments. A computer-readable recording medium may be a tangible device capable of holding and storing instructions for use by an instruction execution device. A computer readable medium can be, for example, but not limited to, an electronic recording device, a magnetic recording device, an optical recording device, an electro-magnetic recording device, a semiconductor recording device, or any suitable combination thereof. Non-limiting examples of computer-readable recording media include the following portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only Memory (EPROM or Flash Memory (registered trademark)), Static Random Access Memory (SRAM), Portable Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD), Memory • sticks, floppy disks, punch cards or mechanically encoded devices having structures protruding into grooves on which instructions are recorded, and any preferred combination thereof. As used herein, computer-readable recording medium includes electromagnetic waves such as radio waves or other freely propagating electromagnetic waves, waveguides or other communication media (e.g., light pulses passing through fiber optic cables); or an electrical signal communicated over a wire, per se, is not interpreted as a transitory signal. In this regard, in various embodiments, computer-readable media as used herein can include tangible, non-transitory computer-readable media.

The computer programs described herein can be downloaded from a computer-readable medium to their respective computing/processing devices, or downloaded to and from, for example, the Internet, local area networks, wide area networks or wireless networks. It can be downloaded to an external computer or external recording device over a network such as a combination. A network may include copper communication cables, optical communication fibers, wireless communications, routers, firewalls, switches, gateway computers and edge servers, or combinations thereof. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and communicates with each computing/processing device to store the computer readable program instructions. into a computer-readable medium within. Computer readable program instructions for performing the operations of one or more embodiments may include assembler instructions, Instruction Set Architecture (ISA) instructions, machine language instructions, machine dependent instructions, micro code, firmware instructions, state Configuration data, configuration data for an integrated circuit, or one or more procedural programming languages, such as Smalltalk®, an object-oriented programming language such as C++, the "C" programming language, or similar programming languages. It can be either source code or object code written in any combination of programming languages. The computer-readable program instructions are distributed entirely on a user computer, partly on a user computer as a stand-alone software package, partly on a user computer, and partly on a remote computer; or run entirely on a remote computer or server. In the latter scenario, the remote computer can be connected to the user computer through any type of network, including a local area network (LAN), wide area network (WAN), or the connection can be an external computer (e.g. through your Internet Service Provider). In some embodiments, an electronic circuit including, for example, a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA) outputs computer readable program instructions and state information of the computer readable program instructions. It can be used to personalize and implement electronic circuitry to carry out one or more aspects of the present embodiments.

One or more aspects described herein may be illustrated in flowchart instructions and/or block diagrams of methods, apparatus (systems), and computer readable media and computer programs according to embodiments of the invention. explained with reference to the program. It is to be understood that any combination of the flowchart illustrations and/or block diagrams and/or the block diagrams in the flowchart illustrations and/or the block diagrams can be implemented by computer readable program instructions. These computer readable program instructions may be provided to a general purpose computer, special purpose computer, or other processor or other programmable data processing apparatus to produce a machine, processor or other machine of a computer. Execution by a programmable data processing device of produces the means for implementing the functions/acts identified in the flowchart and block diagram block or blocks or combinations thereof. These computer readable program instructions, which direct computers, programmable data processing devices and other devices, or combinations thereof, to function in a particular manner may also be stored on computer readable media, The computer-readable recording medium having instructions stored therein constitutes an article of manufacture containing instructions that implement the functional/operational features identified in the block or blocks of the flowchart and block diagrams, or combinations thereof. Computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device, and computer-implemented to a series of operational steps on the computer, other programmable apparatus, or other device. A process causes a computer, other programmable apparatus, or other device to implement the functions/acts identified in the block or blocks of the flowchart illustrations and block diagrams, or combinations thereof.

The flowcharts and block diagrams in the Figures illustrate the architecture, functionality, and possible implementation operations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, the flowcharts or block diagrams can represent modules, segments, or portions of instructions that represent one or more executables for implementing a particular logical function (or functions). commands. In some alternative implementations, the functions noted in the blocks may be performed other than as shown. For example, two blocks shown in succession can be executed substantially concurrently, or blocks can sometimes be executed in the reverse order, depending on the functionality involved. Also, the block diagrams and flowchart illustrations, or both and the blocks and flowchart illustrations in the block diagrams, or combinations thereof, may perform the specified functions or operations, or implement hardware and computer instructions for particular purposes. We point out that it can be implemented by special-purpose hardware-based systems.

Although the subject matter of the present invention has been described above in the general context of computer-executable instructions for a computer program product running on one or more computers, or combinations thereof, those skilled in the art will appreciate that the present disclosure It will also be recognized that it can be implemented in combination with other program modules. Generally, program modules may include routines, programs, components, data structures, etc. that either perform particular tasks or implement particular abstract data types, or both. Furthermore, those skilled in the art will appreciate that the inventive computer-implemented method may be applied to single-processor or multi-processor computer systems, mini-computing devices, mainframe computers in addition to computers, handheld computing devices (e.g., PDAs, telephones, etc.). ), microprocessor-based or programmable consumer, or industrial electronics, and other computer system configurations. The illustrated aspects can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. However, some, if not all aspects of this disclosure can be practiced on stand-alone computers. In a distributed computing environment, program modules may be located in local or remote memory storage devices. For example, in one or more embodiments, computer-executable components may execute from a memory that includes or constitutes one or more distributed memory units. The terms "memory" and "memory unit" as used herein are interchangeable. Moreover, one or more embodiments herein can be computer-executed in a distributed fashion, e.g., by multiple processors executing code from one or more distributed memory units in combination or cooperatively. Able to execute code of possible components. The term "memory" as used herein can cover a single memory at a single location or multiple memories or memory units at one or more locations.

As used in this application, the terms "component," "system," "platform," "interface," etc. relate to a computer-related entity or machine that performs one or more specific functions. Entities may be referenced or contained, or both. The entities disclosed herein can be either hardware, a combination of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, processor, object, executable, thread of execution, program or computer, or any combination thereof. For purposes of illustration, both an application running on a server and the server can be components. One or more components can reside within a process or thread of execution or both, and a component can be localized on one computer or distributed between two or more computers. or both. In alternative embodiments, the respective components execute from various computer readable media having various data structures stored thereon. A component may be a signal carrying one or more data packets (e.g., another component interacting with one component within a local system, a distributed system, or a network with other systems via a signal such as the Internet). , or both) through local or remote processes. As another example, a component may be a device having a specific function by means of a mechanical part operated by electrical or electronic circuitry, which is operated by software or firmware executed by a processor. In such cases, the processor may be internal or external to the device and may execute at least part of a software or firmware application. As yet another example, a component may be a device that provides a particular electronic component without mechanical parts, where the electronic component is a processor or software provided with at least part of the functionality of the electronic component. Other means for executing firmware may be included. In aspects, the components may emulate electronic components via virtual machines within, for example, a cloud computing system.

As used herein, the term "facilitate" means to "facilitate" one or more activities and operations with respect to the nature of a complex computing environment in the context of a system, device, or component. , where multiple components or multiple devices or combinations thereof may be involved in some computing operations. Non-limiting examples of activities that may or may not involve multiple components or multiple devices or combinations thereof include sending or receiving data, establishing connections between devices, intermediate including determining outcomes (including, for example, using machine learning and execution intelligence to determine intermediate outcomes); In this regard, a computing device or component may facilitate operations by manipulating any portion in accomplishing the operations. In the case of the operation of the components described herein, and thus this is described as the operation facilitated by the component, the operation is not limited to: sensors, antennas, audio, or can be completed manually in collaboration with one or more other computing devices, such as visual output devices, other devices, and the like.

Additionally, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless specified otherwise or clear from context, "X utilizes A or B" is intended to mean any of the natural inclusive permutations. That is, if X utilizes A; X utilizes B; or X utilizes both A and B, then "X utilizes A or B" means any Satisfied under things. Furthermore, as used in the subject specification and accompanying drawings, the articles "a" and "an" generally refer to "one" unless otherwise specified or clear from the context to be singular. shall be construed to mean “one or more”. As used herein, the terms "example" or "exemplary" or both are used to mean serving as an example, example, or illustration. For the avoidance of doubt, the instrumentalities disclosed herein are not limited by such examples. Additionally, any aspect or design described herein as "example" or "exemplary" or both is to be construed as preferred or advantageous over other aspects or designs. is not meant to exclude equivalent exemplary structures and techniques known to those skilled in the art.

As used in the subject specification, the term "processor" includes, but is not limited to, a single-core processor; a single processor with software multi-thread execution capability; multicore processors with hardware multithreading technology; and parallel platforms with distributed shared memory. Additionally, the processor may be an integrated circuit, an application specific integrated circuit (ASIC), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic controller (PLC), a complex programmable Reference may be made to logic devices (CPLDs), discrete gate or transistor logic, discrete hardware components or any combination thereof designed to perform the functions described herein. In addition, the processor implements nano-scale architectures such as, but not limited to, molecular and quantum dot based transistors, switches, and gates to optimize space utilization or improve user-mounted performance. can do. A processor may also be implemented as a combination of computing processing units.
In this disclosure, the terms "storage", "record", "data store", "data storage", "database" and substantially any other information storage component related to the operation and functionality of the component are: A "memory component", used to refer to an entity embodied in "memory" or a component that contains memory. The memory and/or memory components described herein can be either volatile memory or non-volatile memory, or can include both volatile and non-volatile memory. By way of example, and not limitation, non-volatile memory can be read-only memory (ROM), electrically programmable ROM (PROM), electrically erasable ROM (EEPROM), flash memory, or non-volatile memory. Random Access Memory (RAM) (eg, Ferroelectric RAM (FeRAM)). Volatile memory can include RAM, which can act as external cache memory, for example. By way of example and not limitation, RAM may include Synchronous RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), SyncLink DRAM. (SLDRAM), Direct Rambus RAM (DRRAM), Direct Rambus Dynamic RAM (DRDRAM), and Rambus Dynamic RAM (RDRAM). Additionally, the disclosed memory components of systems or computer-implemented methods herein are intended to include, without being limited to, these and any other suitable types of memory.

What has been described above includes only examples of systems and computer-implemented methods.
Of course, it is not possible to describe all possible combinations of components or computer-implemented methods for the purposes of describing this disclosure, but many further combinations and permutations of this disclosure are possible to those skilled in the art. It is possible to recognize that there is Further, to the extent the terms "include,""have,""own," etc. are used in the detailed description, claims, appendix and drawings, such term "including" It is intended to be inclusive in a manner analogous to the term "comprise" as it is interpreted when used as a replacement word in the claims.

The description of various embodiments has been presented for purposes of illustration, but is not intended to be exhaustive or limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope of the described embodiments. The terms used herein are used to best describe the principles, practical applications, or technical improvements of the embodiments over those found on the market, or to allow those skilled in the art to understand the embodiments disclosed herein. It has been chosen so that others can understand it.

Claims (20)

a system,
a memory storing computer-executable components;
a processor executing the computer-executable component stored in the memory, the computer-executable component comprising:
a heuristics-based screening component that evaluates a set of artificial intelligence (AI)-designed molecules using one or more classifiers to select a first subset of said AI-designed molecules as candidate drugs; and evaluating said drug candidate using one or more computer simulations of molecular interactions between said drug candidate and one or more biological targets for wet laboratory testing. a simulation-based screening component that selects the second subset of drug candidates of The one or more classifiers classify the AI-designed molecule as having or not having one or more defined characteristics of a target drug based on the molecular sequence of the AI-designed molecule. 2. The system of claim 1, comprising one or more machine learning models for classifying. 3. The system of claim 2, wherein the heuristics-based screening component selects a first subset based on the first subset having the one or more defined characteristics. 2. The system of claim 1, wherein said one or more computer simulations use one or more force field models for said drug candidate and said one or more biological targets. The simulation-based screening component selects the second subset based on a second subset indicative of one or more target molecule interactions in the one or more computer simulations. Item 1. The system according to Item 1. The candidate drug comprises a candidate antimicrobial drug, and the one or more classifiers are such that the AI-designed molecule is antimicrobial peptide, broad spectrum antimicrobial, non-toxic, or structured 2. The system of claim 1, wherein determining if at least one of: The simulation-based screening component uses the one or more computer simulations to determine the relationship between the candidate antimicrobial agent, a model lipid bilayer or another cellular component of a pathogen, and a force field. 7. The system of claim 6, assessing the tendency of the interaction of . The simulation-based screening component selects a second subset of the candidate antimicrobial agents for laboratory testing based on the second subset exhibiting a defined level of propensity for interaction. Item 8. The system according to item 7. The simulation-based screening component uses initial computer simulation to simulate interactions between test molecules having active and inactive sequences and model lipid bilayers or other cellular components of pathogens. , selecting one or more features that correlate to antimicrobial activity based on said interaction. The simulation-based screening component is based on whether the candidate antimicrobial exhibits the one or more characteristics as determined using the one or more computer simulations. to evaluate the candidate antimicrobial agents for inclusion in the second subset. The wet laboratory test is
6. Testing said second subset against one or more pathogens including Gram-positive and Gram-negative bacteria; or Testing virulence of said second subset. The system described in . a method,
classifying said artificial intelligence (AI)-designed molecules from a set of AI-designed molecules as candidate drugs based on classification of said AI-designed molecules using one or more classifiers, by a system operatively coupled to a processor; selecting a first subset; and by said system using one or more computer simulations, molecular interactions between said drug candidate and one or more biological targets. selecting a second subset of said drug candidates for wet laboratory testing based on the evaluation of The one or more classifiers classify the AI-designed molecule as having or not having one or more defined characteristics of a target drug based on the molecular sequence of the AI-designed molecule. 13. The method of claim 12, comprising one or more machine learning models for classifying. 14. The method of claim 13, wherein selecting the first subset comprises selecting the first subset based on the first subset having the one or more defined characteristics. Method. selecting the second subset based on the second subset characterizing one or more target molecule interactions in the one or more computer simulations; 13. The method of claim 12, comprising selecting. The candidate drug comprises a candidate antimicrobial agent, and the classification comprises antimicrobial function, broad-spectrum efficacy, non-toxicity, or presence of defined secondary structure by the system. 13. The method of claim 12, comprising determining whether it includes at least one or more features selected from a group. The method further comprises:
The system uses the one or more computer simulations to characterize interactions between the candidate antimicrobial agent, a model lipid bilayer or another cellular component of the pathogen, and a force field. evaluating, wherein selecting the second subset is based on the second subset indicating a defined level of propensity for interaction. 17. The method of claim 16, comprising: moreover,
The system uses initial computer simulations to assess the interaction between a test molecule with active and inactive sequences, a model lipid bilayer or another cellular component of a pathogen, and a force field. ,
selecting, by the system, one or more features derived from the interaction that correlate to antimicrobial activity;
The system, based on whether the candidate antimicrobial agent exhibits the one or more characteristics as determined using the one or more computer simulations, 17. The method of claim 16, comprising evaluating the candidate antimicrobial agents for inclusion in the subset. The wet laboratory test is
16. comprising at least one of testing said second subset against one or more pathogens, including Gram-positive and Gram-negative bacteria, or testing virulence of said second subset. The method described in . A computer program product for filtering and validating artificial intelligence (AI) design molecules, said computer program product comprising a computer readable storage medium having program instructions embodied therein, said program instructions is executed by the processing component, and the processing component
selecting a first subset of said AI-designed molecules from candidate drugs based on a classification of said AI-designed molecules using one or more classifiers; and using one or more computer simulations. selecting a second subset of said drug candidates for wet laboratory testing based on evaluation of molecular interactions between said drug candidates and one or more biological targets. How to include.

Images